Abstract: Seeks to defend a formal (e.g. truth-conditional) approach to semantic theorizing from advocates of so-called ‘dual pragmatics’ (e.g. relevance theorists or contextualists). I argue, first, that formal semantics is preferable to pragmatically saturated approaches since only formal accounts are compatible with a modularity view of linguistic understanding (a view which is itself, I suggest, independently well motivated). Secondly, I contend that the arguments currently on the table in favour of dual pragmatics fail to show that formal semantics should be abandoned. These arguments for dual pragmatics stem from the existence of overt context-sensitivity in natural languages (for instance, in the form of indexicals and demonstratives) and the purported existence of covert context-sensitivity (in the form of ‘unarticulated constituents’). I look at these arguments in detail and suggest how the formal theorist can accommodate the apparent intrusion of pragmatics into the semantic realm. My defence of formal semantics holds, however, only given an appreciation of what I deem to be the limits of semantic theorizing. Thus, I argue overall for a minimal conception of the nature of semantic theories.

Keywords: context-sensitivity, contextualism, demonstratives, formal semantics, modularity, pragmatics, relevance theory, truth-conditions, unarticulated constituents